Deep Research with Opus 4.6, memory in Model Council, and improved history search

02/12/26

Deep Research with Opus 4.6

Perplexity Deep Research now runs on Opus 4.6, further improving the state-of-the-art results on internal and external benchmarks that we announced just last week (including the Deepmind Deep Research QA benchmark, shown below). Available now for Max users, and gradually rolling out to Pro users. Try using Deep Research to:

Improve SEO strategy: What are the top trending topics in [industry] over the last 90 days? Show me the top-performing content pieces from competitors, the search intent behind trending keywords, and content gaps where competitors aren't publishing. Create a content calendar recommendation for the next quarter based on search volume trends and seasonal patterns.
Compare products in detail: How does the Oura Ring Gen 3 compare to the Samsung Galaxy Ring and the new Ultrahuman Ring Air in terms of sleep tracking accuracy, HRV measurement reliability, battery life, and subscription costs?
Health research: What is the evidence on peer support interventions for chronic pain management, including outcomes on opioid reduction, quality of life, healthcare utilization, and which delivery models work best in primary care settings?

Memory in Model Council

We've extended memory to work with Model Council. Perplexity can now pull in the relevant personalized context for each model, enabling you to get the best answer possible. Try it by asking Model Council to help with decisions such as:

What’s a good training plan for my marathon in May?
What are the top MBA programs I should think about?
What retirement portfolio allocation strategies should I consider given current market?

Improved history search

Finding past threads is now easier—even when you don’t remember the exact words to search for or when the information is buried deep in a conversation. The results include text previews of the matching content, and clicking a thread takes you directly to the exact follow-up where that match appears.

Revamped stock graphs in answers on Android

Searching a stock (such as “nvidia stock”) in the Perplexity Android app now displays an inline stock chart with expanded information and revamped controls.

Kimi K2.5 Thinking

We recently added Kimi K2.5, a new state-of-the-art open source reasoning model from Moonshot AI. We host Kimi K2.5 on Perplexity’s own inference stack in the US, giving us tighter control over latency, reliability, and security for users.